Landing copy density: pain-to-outcome hero, proof chips, cases - #33
Open
shi275773124 wants to merge 88 commits into
Open
Landing copy density: pain-to-outcome hero, proof chips, cases#33shi275773124 wants to merge 88 commits into
shi275773124 wants to merge 88 commits into
Conversation
…cleanup) (#1) * docs: lead with the Sharpe-4 quant audit case; fix badge; drop wallet from README - Headline 'Real case' now leads with the fictional-horizon quant audit (Sharpe 4.06 -> NOT_VIABLE, almost went live) linking the full sanitized log; demote the 12-venue pricing-error case to a collapsible reproducible proof point. - Fix CI badge URL: obsidian-dual-agent -> Falsify. - Remove the EVM wallet line from README Support (FUNDING.yml already renders the Sponsor button). - Mirrored across README.md and README.zh-CN.md. * chore: finish Falsify rename — update stray obsidian-dual-agent URLs Repo-wide cleanup of links that still pointed at the old repo name (redirect-only): demo-vault docs, setup curl URLs (en+zh), pyproject Homepage, the github-action pip-install line, and the link-check self-exclude pattern.
- Hero second line: replace the now-false 'OpenAI won't let you use Claude as the reviewer / cross-vendor review is yours alone' claim (OpenAI ships codex-plugin-cc) with the real, surviving wedge: zero API key (ride the subscriptions you already pay for) + a neutral referee not owned by either contestant, reviewing judgment not just diffs. - Reframe 'One command' into 'Two ways to run it': vault mode (zero key, subscription-authed agents over a Git-synced folder, 3 steps) leads; CLI mode (one command, provider key) is the second option. - Surfaces the keyless path that was previously buried under demo-vault, which is what users actually ask for. - Mirrored across README.md and README.zh-CN.md.
) Add no-API-key backend: `falsify review -p claude` (also codex / gemini / hermes, or any agent via FALSIFY_<NAME>_CMD / FALSIFY_AGENT_CMD) shells out to a locally-installed agent CLI that's already authenticated to its own subscription. Prompt goes in on stdin, stdout is the response — so it rides the subscription you already pay for instead of a per-token API key. - New AGENT_CLIS table + agent_for/agent_cmd/run_agent_cli + a single llm() dispatcher that routes review/draft/run to either the agent CLI or the existing OpenAI-compatible HTTP path. - Any provider name with a FALSIFY_<NAME>_CMD override is treated as an agent, so any vendor's tool wires in — not just the built-ins. - Clear errors: missing CLI on PATH -> actionable message; non-zero exit / empty output surfaced. --dry-run prints the resolved command without running it. - Docs: README 'CLI mode' now notes the keyless -p claude path (en + zh). Verified: dry-run for claude/codex/gemini/hermes + generic override; HTTP preset dry-run unchanged; missing-CLI -> exit 3; lint unaffected; bumped to 0.4.0.
* fix: harden verdict parsing and add tests * fix: nonce draft delimiters + resolve role identity before independence warning - review_prompt now fences drafts with a per-call random tag, so a literal <<<END FALSIFY_DRAFT>>> planted in a draft cannot forge the closing fence; --against reuses the same helper for both versions - role_identity compares the resolved identity (agent argv, or HTTP base+model) instead of raw flags, so '-p deepseek' vs '-p deepseek -m deepseek-chat' correctly trips the author == reviewer warning - tests: forge-resistance + resolved-identity warning cases; drop redundant sys.exit monkeypatches; skip version test when not pip-installed - docs: update bypass model and README wording to match https://claude.ai/code/session_01HCLZidx1giGkS9ShSgQYNv * docs: sync Chinese README with run changes --------- Co-authored-by: shi275773124 <shi275773124@users.noreply.github.com> Co-authored-by: Claude <noreply@anthropic.com>
Combines fff8316 (Windows) + e423ee2/d47c989 (VPS/remote): - templates/risk-scalpel-decision.md: fill-in decision template (from fff8316, not in PR#5) - examples/risk-scalpel-overfit-strategy.md: worked example, high-Sharpe → BLOCK (from fff8316) - README/README.zh-CN: Risk Scalpel hero lead + template/example links + roadmap checkbox - web/serve.py: three-layer cards + updated subtitle (from fff8316 UI additions) - Keeps all d47c989 code hardening (verdict parsing, XSS fixes, tests, CI) - Keeps e423ee2 docs/06-risk-scalpel.md (longer/more complete doc content) Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…ontact info - Update README/README.zh-CN contact links (X: @aishikejian, email) - Add PASS/PASS_WITH_DEBT/BLOCK verdict system to CLI and reviewer - Add local fixture demo mode (no API key required) - Add new docs: getting-started, audit-channel-risks, examples, brooks-lint - Add/update tests for prompt, verdict, Cutline, web markers, demo behavior - Update pyproject.toml metadata and build config Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Document OSS vs paid boundaries for Open Core and add Team edition scope guardrails. Align EN/ZH README messaging around Falsify as a decision gate to keep external narrative consistent. Co-authored-by: Cursor <cursoragent@cursor.com>
Add practical rollout documentation for team-facing delivery, pricing, and adoption metrics. Document BYOK secret setup and policy.yml usage so repositories can enable CI gates safely. Co-authored-by: Cursor <cursoragent@cursor.com>
Add a policy.yml template and a GitHub Action prototype for PR-scoped Falsify gates. This establishes BYOK-friendly rollout defaults with JSON/Markdown artifacts and enforceable BLOCK behavior. Co-authored-by: Cursor <cursoragent@cursor.com>
Add falsify review --json with schema-tagged findings and metadata for CI integration. Introduce strict Known Debt trigger validation that downgrades verdicts to BLOCK when upgrade triggers are missing. Co-authored-by: Cursor <cursoragent@cursor.com>
Refresh the website into a product-led layout with sample/live workbench flows, pricing, and docs browsing routes. Standardize Layer 3 naming to Cutline across docs/templates and extend web tests for i18n and static docs safety. Co-authored-by: Cursor <cursoragent@cursor.com>
Refine EN/ZH opening copy to a sharper product-first hook for GitHub discovery. Keep positioning consistent around Falsify as a decision gate grounded in evidence. Co-authored-by: Cursor <cursoragent@cursor.com>
Focus hero CTAs on GitHub Action install and sample report preview. Add evidence metrics, deliverable artifacts, founder attribution with avatar, and clearer pricing limits. Co-authored-by: Cursor <cursoragent@cursor.com>
Add a practical PR gate setup guide with verification checklist and troubleshooting. Link the guide from README rollout sections so the homepage install CTA has a clear next step. Co-authored-by: Cursor <cursoragent@cursor.com>
Route hero, pricing, and start-section buttons to the 5-minute install doc instead of the raw workflow template. Co-authored-by: Cursor <cursoragent@cursor.com>
Update required marker assertions to match the decision-gate positioning copy introduced in README refresh. Co-authored-by: Cursor <cursoragent@cursor.com>
Rename falsify CI job to test-suite, add concurrency, and gate v* tags with a dedicated release workflow. Document branch protection expectations and the v0.1.0 test-red postmortem. Co-authored-by: Cursor <cursoragent@cursor.com>
Replace the bare docs index with a dark themed layout, sidebar navigation, and richer markdown rendering. Fix active nav state, duplicate CI doc exclusion, and serve_static current_path wiring. Co-authored-by: Cursor <cursoragent@cursor.com>
Prevent regression to generic CTO cite and missing GitHub avatar on the founder quote block. Co-authored-by: Cursor <cursoragent@cursor.com>
Replace GitHub avatar URL with assets/chris-shi-founder.png and bilingual alt text. Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Replace the simple PR comment card with a Checks UI mock that mirrors the falsify-pr-review workflow output format from the Action template. Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Match README.md and site footer: x.com/aishikejian, chrisshi168@icloud.com. Sources: README.md, web/templates/home.html, .github/FUNDING.yml. Co-authored-by: Cursor <cursoragent@cursor.com>
fix(web): mobile homepage cards, trust chips, layer copy
Soften diff-review contrast, clarify BYOK (no Falsify key), label Sharpe case as illustrative with examples/ link, add open-core Known Debt, remove avatar-initial and unify cite as Chris Shi. Co-authored-by: Cursor <cursoragent@cursor.com>
Add docs/self-falsify-2026-06-25.md (Brooks-Lint + adversarial + Cutline, separate Website/Repo verdicts). Soften falsify.py module docstring to match BYOK boundary; update teardown doc to drop testimonial pattern. Co-authored-by: Cursor <cursoragent@cursor.com>
fix(web): self-audit copy + quote attribution
Rebuild mobile scroll as image/card/gap rhythm inspired by CodeRabbit stat strips and Factory gate panels: horizontal proof chips, layers carousel, mini BLOCK verdict panel, case left-stripe badges, accent antipatterns, and larger head-cropped quote avatar. Desktop >=1024px unchanged. Co-authored-by: Cursor <cursoragent@cursor.com>
fix(web): mobile scroll redesign — visual rhythm over text stacks
Chris review pass: correct 审找/门 PASS/rebalance mechanic and align 框架审计/对抗审查 terminology across homepage T.zh, docs, and case copy. Co-authored-by: Cursor <cursoragent@cursor.com>
fix(i18n): zh-CN copy review fixes
Case cards used small in-page /examples/ anchors with a tiny tap target; on mobile the collapsed nav overlay could also intercept clicks. Whole cards now open the repo artifacts on GitHub with external-link affordance, and the mobile nav closes on link tap with visibility hidden when collapsed. Co-authored-by: Cursor <cursoragent@cursor.com>
…workbench CTA Resolves Chris self-Falsify BLOCK findings: protocol verdict copy without NOT_VIABLE/CAUGHT badges, honest proof strip, Sharpe story elevated with deploy deduped to artifact section, expanded workbench with Run sample primary CTA, GitHub blob case links, and updated tests. Co-authored-by: Cursor <cursoragent@cursor.com>
fix(web): landing self-audit BLOCK — verdict coherence, Sharpe hero, workbench
- Remove English/Chinese mixing in 4 doc files - Fix awkward direct translations (reviewer→审议者, silent failure→静默失败) - Normalize Falsify terminology with Chinese annotations - web: clarify run cadence copy Self-Falsify audit: PASS_WITH_DEBT - 0 Must Fix (all 8 findings are pre-existing, not introduced) - 5 Known Debt | 3 Delete - Audit record: docs/self-falsify-2026-06-25-zh-copy-fix.md
fix: polish Chinese documentation copy
* fix: ?lang=zh URL param priority, CN copy cleanup, mobile layout - URL ?lang= now takes precedence over stored pref and writes back - Rebuilt EN/ZH homepage copy, removed garbled Chinese strings - Mobile 390px: proof stats / trust chips / layers wrap instead of truncate - Updated test assertions to match new copy Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix: nav_menu i18n (Menu→菜单), hero mobile gap - home.html: add data-i18n="nav_menu" to toggle button - home.js: add nav_menu key (en: Menu, zh: 菜单) - home.css @960px: hero-inner min-height:auto + align-items:start, hero-visual min-height:200px, hero-copy padding-top:48px → removes ~300px dead whitespace between text and mockup on phone Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> --------- Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
Match production tokens, layout, and zh strings while keeping commercial sections, skills, and /review workbench behavior. Co-authored-by: Cursor <cursoragent@cursor.com>
Ship four falsify-* skills, refresh README and delivery docs, tighten hero/tokens CSS to match production, and add v5 verification screenshots with a Playwright capture script. Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Show BLOCK check artifact in the hero; keep cockpit markup in compat DOM for tests. Add v6 verification shots and hero_image_first assertion. Co-authored-by: Cursor <cursoragent@cursor.com>
…ion-align Align homepage with production UI, add skills pack and commercial positioning
Sync skills install docs (17-skills IA), homepage Install skill CTA and nav, design comps and verification shots, serve /design/ static route, README and getting-started links. Co-authored-by: Cursor <cursoragent@cursor.com>
…design Homepage skills CTA, docs/17-skills, design route
Add falsify-live-production-gate workflow, stale-panel real case, docs/17-skills index, and 2026-06-29 incident-closure audit briefs. Co-authored-by: Cursor <cursoragent@cursor.com>
…design Live Production Gate pack (incident closure)
…zh default. Chris feedback: keep vibe, increase information punch above fold. zh-CN default; restore proof strip + cases grid; honest Self-Falsify why-now. No NOT_VIABLE / 4 CAUGHT regressions (PR24 discipline).
|
Important Review skippedAuto reviews are disabled on base/target branches other than the default branch. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Plus Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
✨ Finishing Touches🧪 Generate unit tests (beta)
Comment |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Test plan
Made with Cursor